Cool Science

Toyota has been - publicly at least - very confused in trying to explain run away vehicles that accelerate uncontrollably and have been connected to truly tragic accidents. It does not look like loose mats any more. More recent reports in the news appear to blame the pedal itself somehow, while private litigates are suggesting an electronic culprit. Toyota has been more silent on this than they probably should have and is promising to announce the definitive fix. Is this a problem with the japanese corporate culture, as many have said? Maybe, probably. But there's potentially another reason: this could also mean that they don't know.

Seems crazy right - how is this possible? The modern car is a study in complexity and when a failure is sporadic and unpredictable, how do you test if the culprit is mechanical, software or electronic? Even worse, nearly every system in a car contributes to 'making the car go' so that there are thousands of mechanical parts directly or indirectly contributing to accelerating the car. And that's just the mechanical. So something is failing sporadically - and spectacularly - and its only symptom is that the car accelerates. Not good. Analysts are railing on Toyota for reckless outsourcing and cost-cutting that may have contributed to this. So don't feel sorry for Toyota, but let's hope they figure out a way to deal with the complexity and find the culprit soon or that the NHTSA does first.

Compare to NASA who manages complexity on an entirely greater level and scope - every successful shuttle launch is a technical marvel in which millions of parts work together in astounding harmony to lift many thousands of pounds off of our fair earth. The feat can barely be described by words. They make it look too easy. I DO feel sorry for NASA.

Anyway, I meant to build all of this up to reflect on the role of complexity in scientific data, and really took my sweet time doing it. Well with over the counter terrabyte storage and wondrous new instruments, we can collect such vast quantities of data that human inspection of the raw results is utterly impossible. Such a thing would normally have been frowned on - experimentalists have always aspired to elegantly design studies that isolate variables one at a time (and still do). But what is new is that there are more ways of extracting meaningful information from vast quantities of data. I met a scientist who can take an orange and tell you exactly where it was grown because they have a wealth of data on oranges in a data base - another who could take a cod (yech) and tell you where it was caught, all in the space of a few minutes (not counting wrestling the cod I suppose).

These are impressive, but the big prize is to demystify body fluids. There are a lot people working to develop what you and I would most closely recognize as the Star Trek 'tricorder'. You pee in a cup - maybe submit a blood sample too - and a few minutes later, you receive an astonishingly complete medical diagnosis of diseases, conditions, etc. This will probably happen in our lifetimes. You heard it here first.

I tried to think of a really cheesy, clever Star Trek quote to wrap this one up and got writer's block about it. It's monday night and BBT was new and awesome so in honor of Sheldon, I'll leave it to Spock,"Life...is not a dream".